JP4226484B2 - Method and system for improving weak selector signal of mixed raster signal - Google Patents

Description

  The MRC (multiple or mixed raster content) representation of a document is versatile. This provides the ability to represent color images and either color or black and white text. The MRC representation allows multiple “planes” to be used to represent document content. MRC representation is becoming increasingly important in the market. This is already established as a major color fax standard.

  In the MRC representation, an image is represented by one or more planes. A major advantage of the MRC representation of a document is that it provides an effective way to store, send and manipulate large digital color documents. This method takes advantage of the characteristics of the human visual system that the ability to identify small color changes is greatly reduced when there is a high contrast edge. Edge information is usually separated from smoothly changing color information and is encoded in one of the planes called selector planes (if possible with a high resolution of 2 bits per pixel or higher). Following careful separation, the various planes can be independently compressed using standard compression schemes (such as JPEG and G4) with good compression and high quality.

  What is needed is a method and system for fully separating the images into a set of planes, taking full advantage of the MRC representation.

  A method for improving mixed raster compression segmentation utilizes a second process stage that generates an MRC selector plane by operating on a multi-bit selector (GraySel) signal generated by the first stage process. . The first stage method used to generate GraySel can be PDL or scan orientation. The binary selector signal generated by the second stage minimizes the apparent compression noise in the reconstructed image. This second stage of processing is done by knowing the size of the JPEG minimum coding unit (MCU) used to compress the segmented foreground and background planes. The idea is to move the fake (soft) edge away from the actual (hard) edge by as much distance as possible to the point where it will fall from the MCU block boundary. Thus, soft edges that occur between two hard edges are eliminated or relocated to an intermediate point, and soft edges that are between the hard edge and the MCU boundary are moved to that boundary. Since JPEG MCU blocks are compressed independently, transitions at the boundary do not cause "ringing" in the decompressed output.

  The present invention provides a method and system for separating an image signal into a set of image planes. The image signal represents a digitally scanned document. Image planes are suitable for mixed raster content (MRC) representations of digitally scanned documents.

  FIG. 1 shows a general MRC representation. This representation includes four independent planes: foreground (FG), background (BG), selector (SEL), and rendering hint (HINTS). In the most general case, there can be multiple foreground and selector pairs at a higher level. However, in many applications this representation is limited to 3 or 4 planes. Background planes are typically used to store continuous tone information such as pictures and / or smoothly changing background colors. The selector plane normally holds a text image (binary) along with other edge information (for example, a line drawing). The foreground plane typically holds the color of the corresponding text and / or line art. However, the MRC representation specifies only the plane and its associated compression method. It does not limit or enforce the contents of each plane in any other way. Each content of the plane can be appropriately defined by performing an MRC representation.

  The MRC structure further considers the fourth plane used to convey additional information about the content of the document, ie the rendering hint plane. For example, the rendering hint plane can implement an ICC (International Color Consortium) color hint that identifies the best color matching method for various objects on the page.

  The foreground and background planes are defined as two full color (L, a, b) or YCC planes. The selector plane is defined as a binary (1 bit deep) plane. One exemplary MRC representation specifies that the foreground and background are JPEG compressed, and that the selector plane is ITU-G4 compressed (standard group 4 facsimile compression). The rendering hint plane is considered arbitrary, but if it is used, a compression scheme similar to the Lempel-Ziv-Welch scheme can be used for the compression. In general, the foreground, background, selector, and rendering hint planes may all be at different resolutions and they need not maintain the original source input resolution.

  The method of assembling a “segmented” MRC image back from its components (eg, planes) is to “pour” the foreground color over the background plane through the “mask” of the selector plane, thereby This is because the content in front of the background plane is overwritten at the position. In other words, this assembly is accomplished by multiplexing between the foreground information and the background information on a pixel-by-pixel basis based on the binary control signal of the selector plane. For example, when the value of the selector is 1, the foreground content is used, and when it is not (that is, when the value of the selector = 0), the background content is used. The multiplexing operation is repeated for each pixel until all output pixels are defined.

  The main advantage of the MRC representation of a document is that it provides an effective way to store, send and manipulate large digital color documents. This method takes advantage of the characteristics of the human visual system that the ability to identify small color changes is greatly reduced when there are high contrast edges. The edge information is usually separated from the smoothly changing color information and coded in the selector plane (possibly with a resolution higher than one selector sample per source pixel). Following careful separation, the various planes can be independently compressed using standard compression schemes (such as JPEG and G4) with good compression and high quality.

  The segment module takes a full color input image to be segmented, three separate outputs for three MRC planes: foreground plane FG, background plane BG, and selector plane Sel, and some additional signals Is generated. A block diagram of the segment module is shown in FIG. The segment module consists of two stages followed by a segmentation stage 24 followed by an MRC scale and tile tag generation stage 44. The segmentation stage 24 can operate in one of two mutually exclusive modes (scan segmentation 24 or PDL segmentation 26).

  The primary input to the scan segmentation module 24 is Src20. This further utilizes the 8-bit screen magnitude estimation signal SCM22 detailed in Applicant's pending US patent application D / A 3011. Scan segmentation module 24 provides full color (raw or raw) foreground plane output Fgr 30 and background plane output Bgr 28 and Sel output 32 (which may have been supersampled by module 24). Output.

  In PDL mode 40, PDL segmentation module 26 does not use SCM 22, but instead can use signal Rht34 to capture hint information from the PDL interpreter and is encoded into CEF hint plane Hnt38. The output from the PDL segmentation module 26 includes a full color (raw or raw) foreground plane Fgr30 and background plane Bgr28, a binary selector plane Sel32, and a hint plane Hnt38 if there is a hint. As indicated above, the hint plane can be 1 bit or 8 bits deep.

  Src images for scan mode 42 and PDL mode 40 are typically processed differently. Scan segment module 24 relies on the input being descreened. This is not required for clean, noiseless images generated directly from PDL sources.

In the scanning process, it is assumed that the chroma component of the source input image Src20 is subsampled by a factor of 2 in the fast scan direction (x: XCSS). Any color image generated by the segmenter does not use XCSS. If a chroma sample of the source image is accessed, no “tuning” filtering is required. That is, for four sample XCSS quadrants, L ₀ A ₀ L ₁ B ₁ , pixel 0 is L ₀ A ₀ B ₁ , and pixel 1 is L ₁ A ₀ B ₁ .

  The output of the selector plane Sel is binary (1 bit deep), and the packed selector plane Spk packs 2 × 2 binary adjacent selector pixels (4 bits) together.

  In PDL processing, the source input Src20 is assumed to be a full color image, where the chroma channel is typically not subsampled and is therefore the same resolution as the luminance channel.

  In general, the exported foreground, background, and selector plane may all be at different resolutions for the input image. For example, the foreground and background planes are typically downsampled and the selector plane may be upsampled from the original input resolution. The amount of upsampling or downsampling is programmable under software control.

  The MRC scale and tile tag generation module 44 reads the first (raw: unprocessed) background Bgr 28, foreground Fgr 30, selector Sel32, and optionally (if PDL mode only), optionally reads the hint Hnt38. This creates a final color MRC layer, background Bgd 46 and foreground Fgd, by sub-sampling and painting pixels that have not been previously specified in the “hole” or raw (raw) image. In addition, the MRC scale and tile tag generation module 44 generates four associated tile tag signals for the background Ttb50, foreground Ttf52, selector Tts53, and optionally (PDL mode only), and optionally the rendering hint Tth54. A tile tag is one binary bit per tile (or strip) and indicates whether the entire current tile can be omitted. This further reduces the overall file size. Missing tiles are automatically painted to the default color for each plane.

  The scan segmentation module 24 is responsible for performing MRC segmentation on the scanned document into three planes. Inputs to the scan segmentation module 24 include an input color signal Src20 and an 8-bit screen magnitude estimation signal SCM22. The scan segmentation module 24 outputs a full color (raw: unprocessed) foreground plane Fgr28, background plane Bgr30, and selector Sel32 plane.

  A block diagram of the scan segmentation module 24 is shown in FIG. The following is a brief description of the various modules that make up the scan segmentation module 24. The color input signal Src20 is a dependent minimum that searches for the minimum Min and maximum Max color values for dynamic thresholding in a 7 × 7 window centered on the current pixel of interest. Transferred to the maximum module 60.

  The minimum Min 61 and maximum Max 63 values are transferred to the dynamic threshold module 62 and the scanning MRC module 64. The dynamic threshold module 62 further uses the input color image Src20 and the 8-bit screen magnitude estimation signal SCM22. The dynamic threshold module 62 outputs a monochrome 8-bit signal Grr55, whose biased zero crossing represents the position of the edge in the selector plane. Also, the dynamic threshold module 62 is further used to communicate to the scanning MRC separation module 64 on a pixel-by-pixel basis when segmentation is applied, in which case how much additional enhancement is provided. An 8-bit segment enhancement control Enh 59 that communicates what is applied is generated.

  The purpose of the block smoothing unit 56 is to move the weak (sometimes referred to as “false”) edges away from the strong edges and contrast in the foreground and background JPEG minimum coding unit (MCU) blocks. To prevent high transitions. In the absence of any strong edges nearby, weak edges are pushed from the JPEG block to the boundary between nearby blocks. This method eliminates unnecessary abrupt transitions within the JPEG block, thus increasing the overall compression and quality. The output from the block smoothing unit 56 is an 8-bit smoothed Grs57 signal representing the smoothed (filtered) form of the incoming signal Grr55.

  The foreground erosion unit 200 is used to address thin (but not broken) text requirements using linear YCC segmentation. A constant value is pulled from the gray selector, which makes the foreground lighter / eroded. This is done only if a nearby test demonstrates that thinning does not result in a dashed line, as will be more fully detailed below.

  Binary scale unit 66 provides the ability to supersample the smoothed gray selector signal Grs57 derived from the output of block smoothing 56. In the normal 1: 1 mode, the Grs57 signal is thresholded to produce a binary selector plane output Sel32. However, in high quality text and line art reproduction, the selector plane may be supersampled at twice the input resolution (eg, 1200 dpi for 600 dpi input). Supersample processing of the selector signal is performed by doubling the sample processing frequency before thresholding. The resulting higher resolution selector pixel is packed into the packed selector signal Spk 122, with the four adjacent packed at once.

  The mark edge processing module 58 takes the packed high resolution selector output Spk 122 and calculates the number of on and off pixels in a 5 × 5 (high resolution) window centered on the current (low resolution) pixel of interest. Count. The output from the mark edge processing module 58 is a 2-bit signal See142. The See signal 142 is set to 0 (corresponding to a constant background area of 3 × 3) if all input pixels inside the 5 × 5 window are off. Similarly, the See signal 142 is set to 3 if all input pixels inside the window are on (corresponding to a 3 × 3 constant foreground area). In addition, the See output is set to 1 or 2 when the 5 × 5 window is almost background or almost foreground respectively.

  Finally, the scanning MRC separation module 64 captures the full color source signal Src20 to be segmented and the minimum and maximum colors (Min, Max) from the dependent minimum / maximum module 60. Further, the MRC separation module 24 uses the See signal 142 from the mark edge processing module 58 and the segmentation and enhancement signal Enh 59 from the dynamic threshold module 62. The MRC separation module 64 actually produces two full color outputs Fgr24 and Bgr30 as coarse estimates for the foreground and background planes, respectively. Here, the various modules of the scan segmentation module are described below.

  A block diagram of the dependent minimum and maximum module is shown in FIG. The dependent minimum / maximum module 60 inputs the Src signal 20 and examines a 7 × 7 window centered on the pixel of interest 80 to find the maximum L and minimum L pixels, where L is the luminance channel. . The maximum output 68 is the pixel having the maximum L72. The minimum output 70 is the pixel with the minimum L74. The minimum output 70 is the pixel with the minimum L74. The resulting chroma value is therefore dependent on the location where the extreme value was found.

  The operation of the dependent minimum and maximum module 60 is illustrated in FIG. This operation proceeds in two stages. In the first stage, the dependent minimum / maximum module 60 searches the window for the largest 68 and smallest 70 samples of the luminance component L. When the positions of the minimum brightness value 74 and the maximum brightness value 72 are found, they are output together with the chroma components (A, B) at these positions. Even if the Src signal 20 arrives at this module with an X subsampled chroma component, this is the point at which the X chroma subsample processing is interrupted. That is, the maximum and minimum color signals do not have X subsampled chrominance.

  This filtering step is separable. For example, the minimum / maximum of an individual column can be calculated first, and then the final minimum 74 can be calculated by finding the column minimum pixel with minimum L. This means that the step-by-step work required as a window performed over the Src image is to calculate one height 7 column and one width 7 row for both minimum and maximum output. become.

  Referring to FIG. 6, the dynamic threshold module 62 applies adaptive thresholding to the incoming color source signal Src20 to generate a raw signed 8-bit gray selector where the zero crossing represents a selector plane transition. A signal Grr output 114 is generated. A gray selector value ≧ 0 marks a pixel with a selector value of 1 and puts it in the foreground. A gray selector value <0 marks a pixel that is placed in the background. As shown in FIG. 6, the dynamic threshold module 60 includes a pair of dependent minimum / maximum (Min, Max) 90 and 92 from the dependent minimum / maximum module 60 and 8 from the screen estimation module (SEM), respectively. A bit screen magnitude estimation signal Scm22 is used. The dynamic threshold module 62 further generates an 8-bit signal Enh118. The Enh signal 118 is communicated to the scanning MRC separation module 64 to determine how much enhancement is applied when the pixel is placed in the background and / or foreground plane.

  The dynamic threshold module 62 operates with three segmentation modules: dynamic threshold, static threshold, and force against foreground. Static thresholding is applied when the image is smooth (does not change). In this mode, pixels with luminance values greater than or equal to DefaultThr76 are assigned to the background (Grr == 127 = −1), and pixels with luminance values less than DefaultThr are assigned to the foreground (Gr = = 129 = + 1). 127 and 129 (+ −1) represent small magnitude values for Grr114. These represent weak decisions that can be corrected by the subsequent block smoothing module 56 by taking into account the location and polarity of strong decisions that are nearby. Said strong decision is represented by signed Grr magnitude> 1 (encoded value <127 or> 129). Strong decisions are generated only in dynamic threshold mode, and only strong decisions can have non-zero Enh codes. Both static thresholding and the mode of force against the foreground produce only weak decisions.

  In some configurations, the force mode against foreground is enabled for halftone images, and this mode is enabled by setting HTFGScmThr84 to some value less than 256. Whenever Scm22 is equal to or greater than HTFGScmThr84, Grr114 is brought to a minimum foreground value of 129 (= ± 1) and Enh118 is set to zero.

  If the force on the foreground is not prioritized, segmentation switches actively between the generation of weak static threshold decisions and the generation of strong dynamic threshold decisions. The signal EnhEn indicates a strong decision and gates the output of the EhFVsScm function to Enh. The EhFVsScm function uses the screen magnitude estimate Scm as the domain value. If true, EnhEn further selects the signed 8-bit signal GSel as the source for encoding Grr. GSel, described in further detail below, is the primary output of the dot product module. As shown in FIG. 6, if the force on the foreground (HTFGEn) is not preferred, the other two outputs of the dot product unit, Ccc and Cc0 (described below), are tested and the results are ORed with each other. Thus, EnhEn is calculated.

  If it is the output of the Ccc ≧ EhClrConThrVsMin function, EnhEn is enabled. Alternatively, EnhEn is enabled if Cc0 ≧ EhLumConThrVsMax and EhLumConThrVsScm functions are maximum. When the input signals for the EhClrConThrVsMin function and the EhLumConThrVsMax function are the luminance components of the Min and Max signals, the input signal for the EhLumConThrVsScm function is Scm.

及びＹは下記数式４の通りである。

The dot product unit 82 (second block from the upper left in FIG. 6) uses the full color input signal Src20 and the full color minimum 92 and maximum 90 (Min, Max) from the dependent minimum maximum unit. These values represent the extreme values of the luminance and the corresponding chroma values found in the (7 × 7) window centered on the current pixel of interest. The operation of this block is mainly to perform a dot product multiplication of two vectors.
[Equation 1]
GSel = minimum (127, maximum (−127 (<X, Y >>>> 256)))
Here, <X, Y> is a dot product operation between two vectors X and Y [Equation 2]
<X, Y> = (X _L , X _A , X _B ) (Y _L , Y _A , Y _B ) ¹ = X _L Y _L + X _A Y _A + X _B Y _B
Here, X is as shown in Equation 3 below.

And Y are as in Equation 4 below.

The embodiment for further improvement, in the case of L _MN == 0, the luminance component value in equation 4 changes from L _MX / 2 to L _MX / 4. This is a primary attempt to adjust the luminance undershoot that is usually produced by the previous sharpening stage. This helps prevent thin text features from being expanded by the segmentation process. That is, it is expressed by the following formula 5.

The (L, A, B) values in Equation 4 or 5 are the corresponding color components of the incoming signal Src20. The X vector in Equation 3 is the vector difference between the maximum and minimum of (Min, Max). The Y vector in Equation 4 is the incoming signal Src20 minus the minimum and maximum averages. By taking the dot product of these two vectors, the output is perpendicular to the X vector and proportional to the relative distance from the plane that intersects it halfway. {X _A , X _B , Y _L , Y _A , Y _B } and the final output GSel can be negative. The absolute magnitude of the dot product output in Equation 1 is not as important as identifying the zero crossing, so this result is simply divided by 256 (shifted right by 8) to fit into the 8-bit range. Converted back to. (Dot product normalization requires division by the magnitude of the vector). However, the output can still often overflow the 8-bit range (approximately 3 or up to a factor of 1.5 bits), so if this becomes too large, limit the output magnitude to 127 It is necessary to add logic to do. The dot product 82 output is shown in FIG. 6 as a signed 8-bit signal GSel or a gray selector output. In order to limit the size of the dot product multiplier to 8 bits, both the X and Y components can be pre-converted by 1/2 and the final divisor can be changed to 64.

The dot product unit 82 further outputs two 8-bit signals that measure luminance and chroma contrast magnitude. Luminance portion Cc0106 is represented by the first component of vector X.
[Equation 6]
Cc0 = X _L = L _maximum− L _minimum

輝度成分における絶対値は、Ｌが正の範囲［０．．．２５５］に制限されるために無視することができ、最大は常に最小より大きい。 A scalar measure for the overall chroma contrast magnitude Ccc 104 is further generated by adding together the absolute values of the two chroma components of the vector X.

The absolute value of the luminance component is a range where L is positive [0. . . Can be ignored and the maximum is always greater than the minimum.

  The decision logic function 91 in the left part of FIG. 6 manages the switch between default and active segmentation mode. Each function is represented by a small set of (x, y) point pairs that represent piecewise linear functions. For x values less than the first x value, the output is the first y value. For x value> last x value, the output is the last y value.

  The importance of the above logic is that either dot product luminance contrast or dot product chroma contrast must be large enough to operate under active segmentation mode. The chroma contrast must be greater than a function of the minimum brightness found in the (7 × 7) window. Similarly, the brightness contrast must be greater than a function of maximum brightness found in the same window, and moreover greater than a function of screen magnitude Scm. Prior to export, the output of the signed gray selector gated by HTFGEn and ENhEn is encoded as 128 by adding 128 to the unsigned 8-bit signal Grr.

  The purpose of the block smoothing unit 56 is to move weak (aka “false”) edges as far away from strong edges as possible. This increases the pulse width in the JPEG plane (decreases the frequency), thereby reducing the compressed size and compression noise. In the absence of any strong edge blocking this, the weak edge is swept out of the block. If a weak edge is trapped between two strong edges, it will be repositioned halfway between the two. This method also coalesces / removes multiple weak edge transitions that improve compression in all three planes.

  The input to the block smoothing unit 56 is the 8-bit raw gray selector output Gr. (Under either active or default segmentation). The output from the block smoothing unit 56 is an 8-bit smoothed glass signal that represents the smoothed (filtered) form of the input signal Gr.

  The very first stage of the block smoothing unit 56 is Gr. Subtract 128 biases (toggling Ms.) to make 'be a signed number. Next, the range of (Gr.-128) is examined. If this is equal to -1 or 1, it is considered to be a weak edge BG or FG, respectively. Anything greater than -1 or 1 is considered to be either a strong BG or FG edge.

  The block smoothing process includes four passes in a square temporary storage area (Tm [Sm] [sz]) that represents the size of the JPEG minimum coding unit (MCU) for the Bgd and Fgd planes. If Fgd is subsampled by a factor of 4 for JPEC compression, then if the chrominance component is additionally subsampled by a factor of 2, the MCU will be 16 × 16. The required Tmp block is sized to Tmp [64] [64] Implementation constraints require that the block size be smaller, in this case 32x32 would be suitable. This pipeline configuration does not use chroma subsampling when applying 1/4 reduction. Only 32 × 32 blocks are required.

  In the first pass, (Grr-128) pixels are processed from left to right in each row's independent Tmp. Whenever there is a strong edge, the position of the corresponding Tmp is initialized to +/− K (predefined constant == 2 * Tmp−1 dimension). The code is the same as Grr-128. Alternatively, for weak edges, Tmp is written with the Tmp value before the magnitude was reduced by one. Since the first path is moving from left to right, the previous Tmp value is just the Tmp value on the left side. The value before the first value in the row is defined to be zero. Decreasing magnitude by 1 converts +4 to +3, -2 to -1, and 0 to 0. In addition, in this pass, all weak values, accumulations of +1, −1 (WeakCnt) are calculated (only where Grr is weak).

  In the second pass, each row of Tmp turns from right to left and this time the value before the magnitude has decreased is compared with the current value. The value with the largest magnitude is placed at the current position. Again, the previous value for the first value on the right is assumed to be zero. Since the rows of passes 1 and 2 are independent, pass 2 can be interleaved with pass 1. Passes 3 and 4 are the same as pass 2 except that the direction is from top to bottom and from bottom to top.

  After the fourth pass, the final smoothed result is generated by examining Tmp. If the Tmp value has a maximum magnitude (+ K or -K), the value used is the value of the original strong edge from Grr. Alternatively, a weak foreground or background code (128 + 1 or 128-1) is used depending on whether the Tmp value is positive or negative.

  Referring to FIG. 7, the binary scale module 66 converts the 8-bit grayscale selector input Grs120 to the binary selector plane output Sel32. For high quality text and line drawings, the binary selector output Sel32 can be maintained at a higher resolution than Src20. For example, the current design allows the use of twice the resolution in each direction (SEG_Sel_ScaleUp), so for a standard 600 dpi scanner, the output resolution of the binary Sel signal may be 1200 dpi. . The selector logic module is responsible for interpolating the gray selector Grs input to a higher resolution and then generating a binary output Sel32 with the signal as a threshold. A redundant Sel (Spk) packed copy is also generated at Src20 resolution.

  A block diagram of the binary scale module 66 is shown in FIG. The 8-bit gray selector input signal Grs120 is first interpolated in the gray domain by a factor of 2 in each direction using a 2D bilinear interpolation. The output from the bilinear interpolation is passed through a multiplier 126 to select whether to use an interpolated (supersampled) selector or a normal binary one that is the same resolution as Src. Finally, the gray output is converted to binary using the threshold unit 128 so that the selector signal Sel32 is generated.

  Note that for the 2x interpolation factor, the binary Sel output data rate is twice as fast as the source signal Src in each direction. This means that for every 8-bit input Grs, the binary scale module generates four binary output Sel samples. A second packed form (Spk) 122 of the Sel output is generated such that the four binary selector bits are packed together, as shown in FIG. It is important to note that the selector output 140 uses vector notation to indicate a higher output resolution. Although the output is still considered binary (ie, it is assumed that the value is only either 0 or 1), each incoming signal Grs input produces four selector bits at the output (2x Assuming an interpolation factor). The four binary pixels are packed into an 8-bit packed signal Spk 122 as shown above. If the interpolation factor is only 1, all four bits are the same.

  Referring to FIG. 10, the mark edge module 58 takes the form of packed high resolution selector Spk 122 and turns on and off in a 5 × 5 [high resolution] window 155 centered on the current pixel 80 of interest. Count off pixels. The output from the mark edge module 58 is a quaternary signal See142. If all input pixels inside the window are zero (corresponding to a constant background area), the See signal 142 is set to zero. Similarly, the See signal 142 is set to 3 if all input pixels inside the window are on (corresponding to a constant foreground area). Further, the See output 142 is set to either 1 or 2 if the window content is mostly back ground or mostly foreground. See has only 4 values and can be encoded with 2 bits.

  The operation of the mark edge module 58 is shown in FIG. The operation of the unit is as follows. The input to the edge processing module 58 is a packed binary selector signal Spk having the same resolution as Src. Edge processing module 58 maintains a 3 × 3 pixel context window centered on the current pixel of interest (at the original input resolution). Logically, the packed selector (Spk) selector includes four binary selector pixels for each Src resolution pixel in the 3 × 3 pixel window, as shown in FIG. The thicker line represents the original Src resolution corresponding to a 6 × 6 pixel context window in the high resolution domain. However, only the content inside the 5 × 5 high resolution pixel area is used, and the shaded area in FIG. 11 is excluded in the count.

  The 5 × 5 high resolution context is designed to “sense” potential edges close to the current pixel of interest. The window pattern uses the full context of two [high resolution] pixels that extend from the top of the current pixel or one down to the left and to the right. The unique window pattern ensures that any edge associated with a large number of (low resolution) pixels does not overlap with nearby pixels, ie no potential edge location can be detected (ie shared) more than once. Is done. The position of 4 × 4 = 16 possible edges within the current window of interest is further illustrated in FIG.

The mark edge module 58 counts the number of high resolution pixels currently turned on in the 5 × 5 high resolution area. This number can be from 0 to 25. This is mapped from the mark edge module 58 to the output signal See as follows.
Seee = 0 If the 5 × 5 count is 0 (no foreground pixel was found) Seee = 1 count is [1. . . 12] (mostly background pixels) See = 2 count is [13. . . 24] (mostly foreground pixels) See = 3 count is 25 (only foreground pixels found)

  Referring again to FIG. 3, the output signal See is transferred to the scanning MRC separation module 64. The See signal is the original input resolution (typically 600 dpi). The scanning MRC separation module 64 is responsible for dividing the incoming source signal Src into foreground and background planes. This module uses the full color minimum and maximum (Min, Max) output from the dependent minimum maximum module and the marked selector edge count signal See from the mark edge module. Furthermore, the scanning MRC separation module has the ability to enhance edge rise via the segment enhancement control signal Enh from the dynamic threshold module.

  Scan MRC separation module 64 outputs two full-color raw: unprocessed estimates of each of foreground output Fgr30 and background output Bgr28. The tracking module, MRC scale, and tile tag generation module then process Fgr and Bfr to generate further final foreground Fgd and background Bgd outputs, respectively.

  The scan MRC separation module 64 takes a full color source signal Src to be segmented and generates values for one or sometimes both of the Fgr and Bgr outputs. The scanning MRC separation module has special codes of zero intensity and chroma (L = a = b = 0) to indicate empty (unsought) pixels in either the foreground Fgr or background Bgr output. . As this process continues across the page, some foreground and background pixels remain sought after. The MRC scale and tile tag generation module then carefully paints the values for these unsought pixels to keep compression low and prevent additional JPEG ringing artifacts.

である場合にはフォアグラウンドにコピーされ、

である場合にはバックグラウンドにコピーされる。したがって、Ｓｅｅ＝＝０である場合にはフォアグラウンドは未定義としてマークされ、Ｓｅｅ＝＝３である場合にはバックグラウンドが未定義としてマークされる。 The scan MRC separation module 64 uses the value of the selector edge count signal See from the mark edge module to determine whether to copy the enhanced Src pixel to the background, foreground, or both. The determination range is shown in FIG. Basically, the enhanced Src pixel is

Is copied to the foreground,

Is copied to the background. Therefore, the foreground is marked as undefined when See == 0, and the background is marked as undefined when See == 3.

Initially, the enhancement factor Enhf1 is a signal Enh incremented by exactly 1, so the maximum value is 256 instead of 255.
[Equation 12]
Enhf = Enh
[Equation 13]
Enhf1 = Enhf + 1

Next, we define the following two full-color enhanced forms of foreground and background (the purpose of which is described below).
[Formula 14]
enhFG = LIM [Src + (Min−Src) (Enhf1 / 256)]
[Equation 15]
enhBG = LIM [Src + (Max−Src) (Enhf1 / 256)]
Implementation note. If the final Bgd and Fgd outputs are XCss (subsampled by X chroma) or scaled down, enhFG and enhBG may be XCSS.

  In equations 14 and 15, Src is a full color input signal, and Min and Max are the dependent minimum and maximum color outputs from the dependent minimum-maximum module. The limiting function LIM has a range of 8 bits [1. . . 255], which eliminates the special zero code provided to mark undefined pixels. Since Src and Min and Max are full color (L, A, B) vectors, the operation is in 3D space.

In the case where foreground is defined, i.e.
See = {1, 2, or 3}
The output Fgr value is
[Equation 16]
When SEE = {1, 2, 3}, Fgr = enhFG
[Equation 17]
When SEE = 0, Fgr = 0
Is required to be.

  If foreground is not used (ie See = 0), the foreground value is undefined by setting the value to the special code Fgr = 0 according to Equation 16 (for all three components). Marked as. Note: This implementation extends Enhf to a 9-bit representation and its value is incremented by 1 (Enhf1) to allow normalization by 256 rather than 255.

  A rigorous examination of Equation 14 shows that the output foreground Fgr value is interpolated between the current input signal value Src and the minimum Min of the dependent minimum maximum module (in 3D space) by the segment enhancement amount represented by Enhf1. Make it clear. If Enhf = 0, no enhancement is performed and the output is set to the input signal Fgr = Src. This is a normal example as long as there is not enough contrast activity in the (7 × 7) window. If Enhf1 = 256 (maximum enhancement), the output is set to the minimum signal Fgr = Min. This represents an example of a pixel in the immediate vicinity of the edge, where it is advantageous to enhance the edge by painting the foreground as dark as possible, as given by the nearby minimum (0 = black). . In general, however, the segment enhancement Enhf can vary between the two extreme values described above, and the output foreground value is correspondingly weighted between the Src and Min values.

Similarly, when using background in segmentation,
That is, when See = {0, 1, 2}, the output Bgr value is

[Equation 18]
If See = {0, 1, 2}, Bgr = enhBG
[Equation 19]
When Seee = 3, Bgr = 0
It is calculated by.

  As before, the output Bgr value varies between the input Src and Max values in proportion to the segment enhancement amount Enhf1, as given by Equation 15. Formula 18 is the same as Formula 16 except that the maximum Max is used instead of the minimum Min and the range of See is different. Using Max for the Bgr output becomes brighter, not darker, as with the foreground.

  Furthermore, when each of the background or foreground is not used (that is, See = 3), as shown by Equation 19 and corresponding to Equation 17, each (see = 0) or background value is By setting the value to the special code Bgr = 0 (for all three components), it is marked as undefined.

  The output from the MRC separation module is two partially painted full color planes Fgr and Bgr. Beyond the selector plane edge, depending on whether it is bright or dark, typically only one of the foreground or background outputs will contain the [enhanced] color of the current pixel. However, nearby edge information may be supported simultaneously on both foreground and background channels.

  Referring to FIG. 2, the PDL segmentation module 26 is responsible for performing MRC segmentation on planes in three PDL documents. The input to the PDL MRC segmentation module 26 can include the input color signal Src20 and any rendering hint Rht 34 that can be provided by the PDL decomposer.

  The PDL MRC segmentation module 26 outputs full color foreground and background planes Fgr28 and Bgr30, a binary selector plane Sel32, preserving some PDL hints, possibly in the 8-bit hint plane Hnt.

  A block diagram of the PDL MRC segmentation module 25 is shown in FIG. Starting from the left, the PDL segmenter reads the input color signal Src20 and the 8-bit rendering hint Rht from the PDL interpreter. The PDL segmenter 26 generates an 8-bit gray selector signal Grr having the same function as that used in the scanning process. Furthermore, the PDL segmenter outputs some PDL hints as MRC hints Hnt.

  The gray selector signal Grr from the PDL MRC segmentation module 25 is processed through a block smoothing unit 56 to produce a smoothed gray selector signal Grs that is forwarded to a binary scale unit 66. The binary scale unit 66 sets the Grs signal as a threshold value so as to generate a binary selector signal Sel. Since the quality of PDL data is not improved by supersample processing, the selector generated by the binary scale unit is always at Src resolution. The operation of the block smoothing unit and the binary scale unit are each described above.

  Finally, the PDL MRC separation module 25 is responsible for dividing the incoming source signal Src20 into foreground and background planes Fgr30 and Bgr28, respectively. This separation is based on the binary selector plane Sel32.

  The PDL segmentation module 26 takes an input color signal Src20 and generates an 8-bit gray selector signal Grr66. In addition, the PDL segmentation module 26 stores some 8-bit PDL interpreter hints Rht as 8-bit PRC hints on the hint plane.

  The operation of the PDL segmentation module is different from the scanning process described above. The scan segmentation process is based on a dependent minimum / maximum analysis followed by dynamic thresholding. However, for clean PDL data, segmentation is based on a classification of pixel content in a 3x3 window centered on the current pixel of interest. This classification is preferred as a set of rules for determining whether the current pixel is associated with a foreground Fgr or background Bgr plane.

  For each incoming Src pixel, the content of the 3 × 3 window around this pixel is analyzed and classified into one or more subsequent classes 158 as shown in the table of FIG. The 3 × 3 window tests are prioritized as shown in the leftmost column in FIG. Thus, for example, a center pixel tagged to be an image pixel by a PDL interpreter takes precedence over any other combination, such as a center pixel tagged as black, white, or text.

  The second column in the table of FIG. 14 displays a list of class names in the C code simulation. The third column gives a brief description of the meaning of the class and how it is tested. Finally, the last column shows how the class is associated (ie segmented) with the foreground or background plane. One exception for the 3 × 3 window test is for classes 6 and 7. The methods for these classes are as follows.

ここで、ＰＤＬＥｑｕａｌＤｉｓｔＬｉｍ及びＰＤＬＯｔｌＤｉｓｔＴｈｒは２つの構成のしきい値である。
２．遭遇する中心ピクセルにＮＥＡＲでなく、さらに該中心ピクセルからＦＡＲである第１のＢａｄではない外側ピクセルは、ＯＴＨＥＲクラスについての基準となる。
３．次に遭遇する中心ピクセルにＮＥＡＲではなく、さらに（該中心ピクセルからＦＡＲではないか、又は、上のＯＴＨＥＲ基準ピクセルにＮＥＡＲではない）ピクセルがＢａｄピクセルとして分類される。
４．最後に、ＯＴＨＥＲクラス６及び７は、遭遇するウィンドウ内のＢａｄではないピクセルに依存する。クラス６（ＯｔｈＤａｒｋ）又は７（ＯｔｈＬｉｔｅ）は基準のＯＴＨＥＲピクセルのカラー値に基づいて差別化される。 1. First, any outer pixel that is not NEAR at the center pixel and is not FAR from the center pixel is classified as a BAD pixel. Here, the meanings of NEAR and FAR are based on the Manhattan distance D _M.
That is,
NEAR if (D _M <PDLEqualDistLim)

Here, PDLEqualDistLim and PDLOtlDistThr are threshold values of two configurations.
2. The first non-Bad outer pixel that is not NEAR to the center pixel encountered and is FAR from the center pixel is the reference for the OTHER class.
3. The next center pixel encountered is not NEAR, and further (not FAR from the center pixel or NEAR to the OTHER reference pixel above) is classified as a Bad pixel.
4). Finally, OTHER classes 6 and 7 depend on non-Bad pixels in the window encountered. Class 6 (OthDark) or 7 (OthLite) is differentiated based on the color value of the reference OTHER pixel.

  The PDL separation module is responsible for dividing the incoming source signal Src into foreground and background planes Fgr and Bgr, respectively. This separation is based on the binary selector plane Sel. The separation process begins by initializing the foreground and background planes with a specially provided zero code (L = a = b = 0) to indicate the unused pixels.

The incoming color Src value is then moved away from zero so that confusion with the “not used” specially provided code is prevented.
Val = Max (1, Src)
The maximum function ensures that Val will never be zero in any of the planes. The separation process continues directly.
When (Sel = 1), Fgd = Val or Bgd = Val

  That is, each incoming color pixel is placed in either the foreground or background. Unlike in the case of scanned documents, this information is not placed in both planes, and is not even near the edges. The separation scheme is therefore greatly simplified for the scanning case.

  Referring to FIG. 15, the MRC scale and tile tag generation module applies additional processing to the coarse foreground and background estimates Fgr and Bgr to generate the final foreground and background outputs Fgd and Bgd. To do. The processing performed by the MRC scale and tile tag generation module is to first subsample the foreground and background values, ignoring undefined pixels. The result is then subsampled by a factor of 8 to calculate the block average (ignoring pixels that are also not defined). The third step is to insert the calculated block average into undefined pixels. The purpose is to reduce artifacts due to JPEG compression ringing by painting undefined pixels with a block average.

  Additional logic inside the MRC scale and tile tag generation module also monitors the foreground level and background output values to detect and flag certain black-only or white-only tiles. Similar logic detects when the selector and hint are zero. A block diagram of the MRC scale and tile tag generation module is shown in FIG.

  All four examples of subsample processing modules operate in a similar manner. The sum of all pixels in the N × N area is calculated and a separate count of the number of pixels where valid = notZero is maintained. This sum is then normalized by a valid pixel count to produce an output. The first stage of sub-sample processing typically spans a 4 × 4 area that represents the entire extent of Fgd and Bgd sub-sample processing. The amount of sub-sample processing is specified by parameters: Seg_Fdg_ScaleDn: SEG_Bgd_ScaleDn: SEG_Fgd_Dst_Css: and SEG_Bgd_Dst_Css.

  The Css parameter controls whether the chroma sample is additionally subsampled by a factor of 2. The second stage always subsamples over an 8 × 8 area that represents the size of the JPEG block at the subsampled resolution. The final normalization of the subsampled output depends on the total weight value. However, it is still possible to avoid division in the formula by using a predetermined multiplication table with a number of options for possible total weight values.

The fill-in block inserts the block averages Fga and Bga into Fgx and Bgx to replace all undefined pixels and generate the final foreground Fgd and background Bgd signals. The hole filling block also produces a very low bandwidth output Tgb / Tgb of 1 bit per tile or strip that can be used to optimize compression when the CEF file is exported. Each fill block monitors each pixel in the tile to test whether all pixels are within the limits set for luminance and chrominance samples. If all tile pixels pass the test, the tile tag bit is set. Three tests are performed on each pixel.
LumRef-L <= TileLumErr {where LumRef is 255 for Bgd and 0 for Fgd}
abs (128-A) <= TileChrmErr
abs (128-B) <= TileChrmErr
The Sel and Hnt tile tag modules are just the equivalent of a large NOR gate operating in one tile block. These produce a 1 if all the binary pixels in the tile are 0. The tile size is programmable, but typical values vary from 64x64 to 512x512 pixels.

  Referring again to FIG. 3, the foreground erosion unit 200 is used to address thin (but not broken) Kanji requirements using linear YCC segmentation. A constant value is pulled from the gray selector, which makes the foreground lighter / eroded. If the pixel is converted from the foreground to the background, this is done only if a nearby test has demonstrated that thinning does not result in a dashed line. Referring to FIG. 16, the figure shows the operation of this module 200. The foreground erosion unit 200 attempts to fit several templates. If a match is found, adjustments are made. FIG. 16 shows two patterns. The shaded block 210 represents the background, and the shaded block 214 represents a stronger foreground that is larger than the adjustment. The shaded block 216 represents a weak foreground that changes to the background when adjustments are drawn. Block 212 is not specified. A weak foreground 216 can switch to the background 210 only if one of the two patterns shown in FIG. 16 is matched (each having four possible orientations).

It is a figure which shows the MRC structure about a document. It is a block diagram of a segment module. FIG. 3 is a block diagram of a scan segment module. It is a block diagram of a dependent minimum maximum module. It is a figure which shows the action | operation of a dependence minimum maximum module. FIG. 4 is a block diagram of a dynamic threshold module. FIG. 3 is a block diagram of a binary scale module. It is a figure which shows the operation | movement of a binary scale module. It is a figure which shows the format of the packed selector. It is a block diagram of a mark edge module. It is a figure which shows the action | operation of a mark edge module. It is a figure which shows the determination range which defines a background and a foreground. FIG. 3 is a block diagram of a PDL MRC segmentation module. It is a table | surface which shows the class of a PDL segmentation module. It is a block diagram of an FG / BG cleanup module. It is a block diagram which shows the erosion of a foreground.

Claims (3)

Improved weak selector signal for mixed raster signals, including foreground signal for foreground plane, background signal for background plane, and selector signal for selector plane that selects one of foreground plane and background plane A way to
(A) A block smoothing unit receives a multi-bit monochrome signal of the mixed raster signal including a selector signal from an image data source device, the monochrome signal indicating a weak selector signal below a reference level and above the reference level. Indicates a strong selector signal ,
(B) the block smoothing unit partitions the selector signal into a plurality of uniform blocks;
(C) the block smoothing unit performs two-dimensional filtering on each block;
(D) the block smoothing unit replaces the weak selector signal with the filtered result;
A method characterized by comprising. The method of claim 1 , further comprising:
(E) the block smoothing unit subtracts a bias corresponding to the reference level from the monochrome signal to make the monochrome signal a number with a sign (+ or-sign);
(F) including that the block smoothing unit determines that the monochrome signal is a weak selector signal when the signed number is +1 or -1.
A method characterized by : Improved weak selector signal for mixed raster signals, including foreground signal for foreground plane, background signal for background plane, and selector signal for selector plane that selects one of foreground plane and background plane A system that
(A) receiving a multi-bit monochrome signal of the mixed raster signal including a selector signal from an image data source device, the monochrome signal indicating a weak selector signal below a reference level and a strong selector signal above the reference level; It is shown
(B) means for partitioning the selector signal into a plurality of uniform blocks;
(C) means for performing two-dimensional filtering on each block;
(D) means for replacing a selector signal weaker than a reference level with the filtered result;
A system characterized by including .

Images